Q3 Revenue $15.8 Billion, Up 12% Year-on-Year… Product Orders Surge 35%
Cisco is re-emerging as a core beneficiary of the AI infrastructure investment cycle. Cisco reported revenue of $15.8 billion in its fiscal Q3 2026, an increase of 12% compared to the same period of the previous year. The company evaluated this performance as "double-digit revenue and profit growth exceeding the upper end of guidance." GAAP EPS was $0.85, up 37% year-on-year, while non-GAAP EPS was $1.06, up 10%.
The core of this performance is not simply revenue growth. The important point is that AI data center investment and enterprise network replacement demand are moving simultaneously. Cisco disclosed that total product orders in Q3 increased 35% year-on-year, and even excluding hyperscalers, increased 19%. In particular, networking product orders increased more than 50% year-on-year, and data center switching orders also grew more than 40%.
This demonstrates that the bottleneck of the AI era does not remain only with GPUs and power. Training and operating massive AI models requires not only servers, storage, and accelerators inside data centers, but also high-performance network infrastructure that connects them quickly and stably. Cisco's performance is an indicator demonstrating that AI infrastructure competition is spreading from semiconductors to networks, security, observability, and campus infrastructure.
A point Cisco particularly emphasized is hyperscaler AI infrastructure orders. The company disclosed that it has secured $5.3 billion in AI infrastructure orders on a fiscal year-to-date basis. More important is the upward revision of forecasts. Cisco raised its FY26 AI infrastructure order expectations from the existing $5 billion to $9 billion, and also raised related revenue expectations from the existing $3 billion to $4 billion.
This is a point that can change the market's interpretation. Until now, the representative beneficiaries of AI infrastructure investment were NVIDIA and cloud operators, and some server and power infrastructure companies. However, Cisco's performance this time means AI investment is beginning to be reflected in earnest in network equipment companies as well. The more computation AI data centers require, the faster and more stable data movement becomes necessary. Competitiveness in the AI era is determined not only by chip performance but also in the network structure that connects chip to chip, server to server, and data center to cloud.
Looking at business segments, the networking segment led performance. Networking revenue in Q3 was $8.815 billion, up 25% year-on-year. Meanwhile, security segment revenue was $2.008 billion, on par with the previous year, and the collaboration segment was $1.024 billion, down 1%. The observability segment was $269 million, up 3%. Total product revenue was $12.117 billion, up 17%, while service revenue was $3.724 billion, down 1%.
Regional performance also improved evenly. The Americas region revenue was $9.569 billion, up 14%; EMEA was $4.054 billion, up 9%; and APJC was $2.218 billion, up 9%. This demonstrates that the recovery is not a rebound limited to specific regions but that global network investment demand is recovering simultaneously.
However, this performance cannot be viewed with pure optimism. Although total revenue increased significantly, margins declined compared to the previous year. GAAP gross margin was 63.6%, down from 65.6% in the same period of the previous year, and non-GAAP gross margin also fell to 66.0% from 68.6%. Product gross margin also declined from 64.4% to 61.9% on a GAAP basis.
This suggests that even with strong AI infrastructure demand, factors such as supply chain costs, product mix, intensifying competition, and tariff impacts can pressure profitability. In fact, Cisco disclosed that its Q4 and FY26 guidance reflects tariff impacts based on current trade policy.
Another noteworthy point is restructuring. Cisco announced a restructuring plan to invest in core growth opportunities such as silicon, optics, security, and AI. In connection with this, it expected to recognize pre-tax costs of up to $1 billion, of which approximately $450 million is projected to be reflected in fiscal Q4 2026.
This means Cisco has entered portfolio restructuring, not simply responding to current demand. Transitioning from a network equipment company to a core infrastructure company of the AI era requires difficult restructuring without maintaining existing organizational and cost structures. AI networking, silicon, optical modules, and security integration require different speeds both technically and financially.
Future prospects have also been revised upward. Cisco presented Q4 FY26 revenue of $16.7 billion to $16.9 billion, and non-GAAP EPS of $1.16 to $1.18. Full FY26 revenue guidance was presented as $62.8 billion to $63.0 billion, and non-GAAP EPS guidance as $4.27 to $4.29.
The industrial message this performance delivers is clear. AI infrastructure investment has now gone beyond the first wave and is entering the second wave. If the first wave was about securing GPUs and cloud computing resources, the second wave is the overall reconstruction of infrastructure including networks, security, data center switching, campus network replacement, and observability.
In particular, the corporate campus network replacement cycle is also important. Cisco disclosed that a multi-year large-scale campus networking refresh cycle is underway, with campus networking orders increasing more than 25% year-on-year. It also stated that the spread speed of the next-generation portfolio is faster than past product launches.
This connects to the fact that AI does not only operate inside data centers. For companies to adopt AI agents, video analysis, security automation, collaborative AI, and edge AI, the networks of offices, factories, stores, and campuses must also change. AI transformation is expanding from a problem inside the cloud to a problem of the entire company's network structure.
Cisco's performance this quarter shows that when AI investment reaches the second wave, the network equipment companies that were on the sidelines of the AI boom can lead the next stage of performance. The recovery of network stocks is not just a cyclical rebound. It reflects the structural reality that AI requires networks, and better AI requires better networks.
